…
# The one adaptive heartbeat (plan 2026-11-17-one-adaptive-heartbeat), 2026-11-17 The FIRST attempt (withdrawn the same day, never committed: the beat on fabric steps, Loop #7 from Hera, the Stadium charged on the beat) left two logs that the second attempt overwrote; what remains here is the second attempt's: - test_heartbeat-before.log: kernel/tests/test_heartbeat.c before the pacing existed: does not build. - test_heartbeat-after.log: 11 checks, 0 failures. - test_stadium_node-before.log / -after.log: stadium_resident_heat_total missing, then 16 checks, 0 failures. - test_mesh_boot-before.log: the lockstep checks before the mesh had a clock: does not build. - test_mesh_boot-after.log: 103 checks, 0 failures, POST 990/990. - build-amd64.log: the amd64 kernel builds clean. - kernel-tests.log: every kernel host test passes. - v4-test.log: make -C v4 test, ended by the 590 s foreground limit inside mesh-boot after both widths' programs had passed (81 programs, 0 failures) -- not a failure; the rest run one by one: v4-mesh-boot.log (103), v4-mesh-boot-on.log (112), v4-mesh-boot-noop.log (100), v4-inference.log (6): 0 failures. - v4-san-mesh-boot.log: the mesh boot under ASan+UBSan: 103 checks, 0 failures. The tree was not yet committed when these ran; the protocol on the committed tree is the acceptance evidence.